Differences in posture kinematics between using a tablet, a laptop, and a desktop computer in sitting and in standing

2.1Study design

This within subject observational study compared the posture of the same individual in four different conditions (sitting with a desktop computer, standing with a desktop computer, sitting with a laptop and sitting with a tablet) using three dimensional (3D) motion analysis.

2.2Participant recruitment

Sample size calculation was based on previous data indicating the standard deviation of neck flexion/extension ranges of motion between 8.1° when sitting at a desk and 9.7° when using a laptop [4]. Using an average estimate of the variation in neck flexion/extension of 9°, a difference in neck flexion/extension angle of 5° between workstations could be detected with 80% power at an alpha of 0.05 with a sample size of 27 per comparison. As this was a within-subjects design, we set the sample size for recruitment at 30.

Volunteers for this study were recruited by advertising the study on noticeboards around the University campus where the study took place and on the University Facebook pages. Participants were required to have had at least some previous experience of using a tablet, a laptop and a desktop computer. Potential participants were excluded due to chronic disease or deformity in the neck, back or arms; or pain, injury or trauma in these areas for which they sought treatment in the past 2 years. This research complied with the American Psychological Association Code of Ethics and was approved by the Institutional Review Board at the University of Newcastle. Australia (H-2015-0039). Informed consent was obtained from each participant.

2.3Task

The most common online activities, email (performed by 97% of internet users), browsing (96%) and searching for information (95%), are related to typing and editing [20]. Therefore, a 5-minute editing task requiring typing and editing was performed on each device. Participants first performed a 2-minute typing task on each device as a warm-up to familiarize themselves with the workstation, the electronic device and the software used. This 2-minute typing task required participants to type out Edgar Allen Poe’s The Raven, and data were not collected during this task. The 5-minute editing task performed in each condition (laptop, tablet, and desktop sitting, desktop standing) involved selecting, deleting and typing new phrases using prompts in a standardised article in Microsoft Word (Microsoft Corp., Redmond, WA, USA), previously used by Gold et al. [12]. Data were collected during the fourth minute of the editing task for each condition. In order to minimise the effects of altered posture due to anticipation or familiarized content in the comparison of workstations, all participants completed each workstation in a randomised order determined by a computerised random number generator.

2.4Electronic devices

The keyboard of each electronic device was positioned 6–7 cm from the edge of the desk, and for the desktop conditions using the external keyboard, it was laid flat without using the two posteriorly placed supporting legs [21, 22]. The 9.7-inch iPad4 with no case or other add-ons was placed on the desk flat and in landscape orientation. For the tablet condition, Microsoft 365 Word was used on iPad 4 with iOS 8 (Apple Inc., Cupertino, CA, USA) operating system to ensure the computerized editing activity was the same on each device. A 15.6-inch laptop (Model No. E6540, Dell Inc., Round Rock, TX, USA) with no mouse or other add-ons was used for the laptop component in this study. The screen tilt angle was adjusted to the participant’s preference, as ideal tilt angles for working on laptops have not been established [23]. For the laptop, Microsoft Office Word Home and Business 2013 (Microsoft Corp., Redmond, WA, USA) was used with the Microsoft Windows 7 operating system. For the simulated desktop conditions (sitting and standing), the same laptop was equipped with a standard 104-key Windows keyboard (Model No. KB212B), a chorded optic mouse (Model No. MS116) and a 23-inch monitor (Model No. P2314 H, Dell, Inc, Round Rock TX, USA) mounted on an adjustable sitting/standing workstation. The display was projected on the monitor using a standard Video Graphics Array (VGA) cable with the laptop lid in the closed position. The devices and their positioning for each condition are illustrated in Fig. 1.

Fig. 1.

Device placement for each experimental condition: (A) tablet, (B) laptop, (C) desktop, sitting, and (D) desktop, standing, with retro-reflective marker configuration.

2.5Condition descriptions

The sitting workstation for the tablet, laptop and desktop computer included a standard 760 mm height desk and a 5-pronged caster base, height adjustable office chair (Model No. QU26, Sturdy Framac, Padstow, NSW, Australia). Chair height was measured in standing with the highest point of the seat right below kneecap. Participants were seated so the back support of the office chair was used at all times when they were in sitting [22].

For the desktop conditions, the flat panel monitor was mounted on an adjustable sitting/standing workstation (Kangaroo Pro, Ergonomic Essentials, North Ryde, Australia) placed on top of the standard office desk. The sitting/standing workstation had an independently adjustable screen height and work surface height. For sitting and standing, the screen height was set to have the top one-third of the screen at or below eye level. For standing work surface height was set to have the top of the desk at elbow height [22]. This adjustable workstation was removed from the desk for the laptop and tablet conditions.

2.6Procedure

Participants attended a university laboratory where they completed a demographic questionnaire including: age, gender, dominant hand, occupation, physical activity using the Godin-Shepherd Leisure-time Physical Activity Questionnaire (GLPAQ) [24], types of electronic devices used every day (desktop computer, laptop computer, tablet, phone), and average time spent on each. Six Qualisys motion capture cameras (100 Hz; 300+, Qualisys AB, Gothenburg, Sweden) were used to record the three-dimensional (3D) movements of the head, trunk and upper limbs. Retro-reflective markers were taped on bony land marks bilaterally on the mastoid, acromioclavicular (AC) joint, greater tubercle of the humerus, anterior and posterior glenohumeral joint, medial and lateral epicondyle, radial and ulnar styloid process, head of the third metacarpal, and laterally on the iliac crest and lowest point on the ribs. Single markers were also taped on the C2 spinous process, sternal notch (SN), and xiphoid process. In addition, five tracking clusters of three markers each were positioned with medical adhesive tape on bilateral humeri, forearms and C7. A tracking headband of four markers was positioned around the head, so that markers aligned vertically with the nose anteriorly, tragus of the ear laterally, and midpoint of the external occipital protuberance posteriorly.

2.7Kinematic analysis and reduction

Three-dimensional kinematic data were analysed using Visual 3D software (Version 5; C-Motion, Germantown, MD, USA). All raw kinematic coordinates were filtered using a fourth-order zero phase-shift Butterworth digital low-pass filter (f_c = 8 Hz) prior to calculation of individual joint kinematics. Geometric primitives were used to model the inertial properties of each segment [25], classifying the neck and thorax as cylindrical [26]; the upper arm and forearm as a frusta of a cone [27]; the hand as spherical; and the head and head-neck as an elliptical cylinder.

Traditionally, the head and neck segments are treated as one segment (head-neck); however, to enable more complex kinematic modelling, separate head and neck segments were created. The distal joint centre of the head was defined as the mid-head landmark (the midpoint of the four headband markers), with a radius of 50% of the distance between the left and right lateral headband markers. The proximal joint centre of the head was defined as the prox-head (the axial and medial lateral position of the mid-head landmark and the axial position of C7-SN landmark (50% distance between these C7 and SN markers) with a radius of 50% of the distance between the left and right C7 markers. The neck joint centres were defined as for the proximal joint as the C7-SN landmark, and distal joint as the C7-SN landmark with a 0.1 m axial offset inferiorly with a radius of 50% of the distance between the left and right C7 markers. The traditional head-neck segment joint centres were defined as the C7-SN and mid-head landmarks, with a radius of 50% of the distance between the left and right lateral headband markers.

Cartesian local coordinate system sign convention was conserved by defining all segments as follows: the x-axis as mediolateral axis, the y-axis as anterior-posterior axis and the z-axis as superior-inferior axis. Intersegmental joint angles were conveyed using an x,y,z Cardan sequence of rotation. Data collected at 100 Hz included joint angles (°) for head-neck (HN) and headneck-trunk (HNT) flexion/extension, lateral flexion (left or right) and rotation (left or right), bilateral elbow flexion, bilateral wrist flexion-extension and radial-ulnar deviation. As a measure of shoulder elevation usually considered as a marker of excessive upper trapezius muscle activity and tension in the upper shoulders and neck, AC-C7 vertical distance (mm) was measured by calculating the mean vertical distance from the acromioclavicular joint marker perpendicular to a horizontal line bisecting C7. To calculate mean bilateral shoulder elevation (mm) during a task, the AC-C7 vertical distance measured in a static standing position was subtracted from the mean AC-C7 vertical distance during the one-minute typing tasks. Although the psychometric properties of posture measurement using 3D motion analysis have not been fully established [28], 3D motion analysis has demonstrated high test-retest reliability for joint angles [29] (the focus of the current study), and greater reliability when using marker ‘clusters’ (as used in the current study) rather than single markers for measuring spine movements [30].

2.8Statistical analysis

Descriptive statistics were used to report participant characteristics and posture variables. Differences in the posture variables between the four workstation conditions were compared using repeated measures analyses of variance (RM ANOVA) with Bonferroni post-hoc tests. The significance threshold was set at (P < 0.05). Each posture variable was examined separately using RM ANOVA, with the repeated factor being workstation (tablet, laptop, desktop [sitting] and desktop [standing]). Partial Eta² are reported for each RM ANOVA to indicate the size of the effect of workstation on posture. Greenhouse-Geisser adjustment was used when Mauchley’s test of sphericity was not met. SPSS v21.0 (IBM Inc., Armonk, NY, USA) was used for all statistical analyses.

3.1Participants

Thirty participants completed the study (17 females, 13 males, mean age 25.9±7.42 years, range 20–50, mean height 171.62±9.22 cm, mean weight 69.90±14.14 kg). Three participants (10%) reported left hand dominance, however all of them reported that they typically used a computer mouse in their right hand; therefore, all participants used a right-handed mouse in this study. Twenty-three (76.7%) were university students and seven (23.3%) were employed full-time in various occupations, e.g. clerical, wait staff. The majority of participants were considered active, scoring≥24 on the GLPAQ (80.0%, n = 24) with 10% (n = 3) moderately active and 10.0% (n = 3) insufficiently active.

Eleven participants (36.7%) self-reported daily desktop computer use, 25 (83.3%) laptop, 16 (53.3%) tablet and all 30 (100%) daily smart phone usage. The daily desktop usage had a mean of 1.45±2.45 hours; laptop 3.18±2.54 hours, tablet 0.70±1.09 hours; and smart phone 3.07±3.61 hours.

3.2Postural variables

Descriptive data for posture variables for each workstation are reported in Table 1. The variables of wrist flexion/extension and radial/ulnar deviation at the tablet station only include 29 participants, as one participant was an outlier; they rested their left arm on table without using it for the editing task.

3.3Effect of workstation

There was a statistically significant effect of workstation on all postural variables, except for HN lateral flexion and HN rotation (Table 1). Therefore, pairwise differences in posture variables between workstations are reported for all variables excluding these two in Tables 2 and 3. Observations of differences between workstations for the HN and the HNT postural variables were similar (Table 1), but overall differences did not reach significance for HN lateral flexion and HN rotation whereas these were significant for the HNT angles. The difference between workstations for AC-C7 vertical distance and shoulder elevation was the same, as the calculation of shoulder elevation represented a simple transformation of the AC-C7 vertical distance data. Therefore workstation differences are only included for shoulder elevation in Tables 2 and 3.

Significantly greater HN and HNT flexion, bilateral elbow flexion, and shoulder elevation occurred when using the tablet compared to all three other workstations, though the difference in right shoulder elevation between the tablet and the laptop did not reach significance (Table 2). There was greater wrist flexion using the tablet compared to the desktop in sitting or standing, but not the laptop (Table 2). Though there was statistically less HNT lateral flexion using the tablet compared to the three other workstations, the magnitudes of the absolute differences were functionally irrelevant (<2°, with 95% CIs for each of the three comparisons spanning approximately 3° or less), and also suggested by a small effect size (0.212).

When comparing the laptop to the desktop computer (in both sitting or standing), laptop use resulted in significantly greater HN, HNT, elbow and wrist flexion and shoulder elevation (Table 3), except for right shoulder elevation between the laptop and the desktop in standing. Participants displayed significantly greater right wrist ulnar deviation using the laptop compared to the desktop in standing, but this difference did not reach significance when comparing the laptop to the desktop in sitting. Most postural variables were similar between a sitting and a standing desktop (Table 3). Although participants displayed significantly greater HNT left rotation when using a desktop in sitting compared to standing, the magnitude of the absolute difference was functionally irrelevant (1.28°, 95% CI 0.23, 2.29, effect size 0.100).

4.1Limitations

The five-minute tasks participants completed in this study may not reflect a real working environment where individuals remain in postures for much longer time durations. The short length of the task may be one possible reason that no differences in postures were found between the sitting and standing desk conditions. A longer duration of study tasks would provide a more realistic work environment and should be considered in future research. Furthermore, different participants have various levels of English proficiency and computer literacy; therefore, the editing task may be very easy for some participants but difficult for others. This may possibly affect posture as the participant struggling with the task becomes more skilled as each subsequent workstation is evaluated. To minimise any learning effects of the task, we randomised the order of workstations for each participant and the within-subject design reduces confounding factors in the posture comparisons between workstations.

In the current study participants used the tablet with it placed flat on a desk rather than a participants preferred position, in order to minimise additional confounding factors between participants from variations in positions. This use of a tablet may not be as common as holding it on one’s lap, or propping it up, and future studies could investigate postural differences between these tablet positions. Postures during the use of smart phone were not included in the current study as the standardised typing task could not be performed on a smart phone. However, postures during smart phone use would be an important addition to future studies investigating the role of posture and neck pain.